James E. Fitzgerald
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
James E. Fitzgerald has a total of 108 co-authors. Here are the top co-authors by number of shared publications:
- Damon A. Clark
Yale University
5 shared publications - Florian Engert
Harvard University
4 shared publications - Maanasa Natrajan
Department of Neurobiology
4 shared publications - Mark J. Schnitzer
CNC Program, Stanford University, Stanford, CA 94305, USA
4 shared publications - Nelson Spruston
Janelia Research Campus, Howard Hughes Medical Institute, Ashburn, VA 20147, USA
4 shared publications Abhishek K. Jha
Department of Biochemistry and Molecular Biology
3 shared publications
Latest Publications
- Open Access
- 1 October 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1) , 10412
Author Correction: Whole-brain, all-optical interrogation of neuronal dynamics underlying gut and vascular interoception in zebrafish
- 1 October 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1) , 10412
Open AccessAuthor Correction: Whole-brain, all-optical interrogation of neuronal dynamics underlying gut and vascular interoception in zebrafish
- Open Access
- 27 August 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Whole-brain, all-optical interrogation of neuronal dynamics underlying gut and vascular interoception in zebrafish
To select behaviors appropriate to their circumstances and needs, animals integrate information about the external environment with information about the state of their bodies, derived from sensing and processing internal signals (interoception). However, the brain-wide circuit activity underlying interoception and its integration with sensorimotor processing remains unclear, partly because of technical barriers to accessing whole-brain activity at the cellular level during physiological perturbations. We developed an all-optical system for whole-brain neuronal imaging in behaving larval zebrafish during optical uncaging of gut- or bloodstream-targeted nutrients and visuo-motor stimulation. Widespread neural activity throughout the brain encoded nutrient delivery, unfolding on multiple timescales across many peripheral and central regions. Evoked activity depended on delivery location and occurred in response to both amino acids and D-glucose, but not L-glucose. Many gut responsive neurons also responded to swimming and visual stimuli, with brainstem areas primarily integrating gut and motor signals and midbrain regions integrating jointly gut, visual, and motor signals. This platform links body-brain communication studies to brain-wide neural computation in awake, behaving vertebrates.
- 27 August 2026
- Published by Springer Nature in Nature Communications
- Vol. 17 (1)
Open AccessWhole-brain, all-optical interrogation of neuronal dynamics underlying gut and vascular interoception in zebrafish
To select behaviors appropriate to their circumstances and needs, animals integrate information about the external environment with information about the state of their bodies, derived from sensing and processing internal signals (interoception). However, the brain-wide circuit activity underlying interoception and its integration with sensorimotor processing remains unclear, partly because of technical barriers to accessing whole-brain activity at the cellular level during physiological perturbations. We developed an all-optical system for whole-brain neuronal imaging in behaving larval zebrafish during optical uncaging of gut- or bloodstream-targeted nutrients and visuo-motor stimulation. Widespread neural activity throughout the brain encoded nutrient delivery, unfolding on multiple timescales across many peripheral and central regions. Evoked activity depended on delivery location and occurred in response to both amino acids and D-glucose, but not L-glucose. Many gut responsive neurons also responded to swimming and visual stimuli, with brainstem areas primarily integrating gut and motor signals and midbrain regions integrating jointly gut, visual, and motor signals. This platform links body-brain communication studies to brain-wide neural computation in awake, behaving vertebrates.
- Open Access
Behavioral alignment as an organizing principle in sensory coding
The ultimate goal of sensory coding is to extract and represent the cues required for adaptive motor output. This suggests that sensory codes and behavioral outcomes may align, and a variety of studies have argued that both biological and engineered sensory systems represent stimuli similarly when they play similar roles in behavior. However, the extent to which behavioral demands determine sensory coding throughout the brain is largely unknown. Here we propose that behavioral alignment is a general principle that organizes sensory representations, and we show that carefully measured behavior can predictively account for visual encoding across the entire zebrafish brain. We discover population codes that represent visual motion stimuli according to the optomotor responses elicited by them, indicating that information required for behavioral selection is explicitly encoded in sensory populations. Brain-like neuronal responses result when sensory codes are optimized for efficiency and aligned to behavior. These results provide a paradigm for understanding sensory representations through the behaviors they drive.
Open AccessBehavioral alignment as an organizing principle in sensory coding
The ultimate goal of sensory coding is to extract and represent the cues required for adaptive motor output. This suggests that sensory codes and behavioral outcomes may align, and a variety of studies have argued that both biological and engineered sensory systems represent stimuli similarly when they play similar roles in behavior. However, the extent to which behavioral demands determine sensory coding throughout the brain is largely unknown. Here we propose that behavioral alignment is a general principle that organizes sensory representations, and we show that carefully measured behavior can predictively account for visual encoding across the entire zebrafish brain. We discover population codes that represent visual motion stimuli according to the optomotor responses elicited by them, indicating that information required for behavioral selection is explicitly encoded in sensory populations. Brain-like neuronal responses result when sensory codes are optimized for efficiency and aligned to behavior. These results provide a paradigm for understanding sensory representations through the behaviors they drive.
- Open Access
Memory from variability: Heritable short-term cellular memory emerges from stochastic biochemical reaction networks
Cells exhibit a mysterious form of selective heritable short-term memory, influencing outcomes as diverse as cell fate decisions in embryos and environmental responses in cancer cells and bacteria. Here, we present a simple theoretical framework explaining how this selective memory can arise from the reactions regulating molecular levels in cells. Our key insight is that related cells retain more similar molecular concentrations relative to random cells when a greater variance of possible concentration states is created during a single cell generation than is created by cell division across a population. This persistence of molecular similarity down a lineage constitutes a form of heritable short-term memory. We identify the biochemical networks that produce, modify, and degrade molecules as an underexplored source of these additional molecular concentration states. Using experimentally informed simulations, we find that the strength and duration of molecular similarity down a lineage depend on tunable network properties, explaining why some cellular traits persist only briefly while others last generations. These contributions to molecular concentration variance from biochemical reaction networks act in concert with gene expression and other regulatory processes to shape the protein composition of cells. Our framework yields clear, testable predictions for determining how biochemical network architectures drive non-genetic cellular inheritance.
Open AccessMemory from variability: Heritable short-term cellular memory emerges from stochastic biochemical reaction networks
Cells exhibit a mysterious form of selective heritable short-term memory, influencing outcomes as diverse as cell fate decisions in embryos and environmental responses in cancer cells and bacteria. Here, we present a simple theoretical framework explaining how this selective memory can arise from the reactions regulating molecular levels in cells. Our key insight is that related cells retain more similar molecular concentrations relative to random cells when a greater variance of possible concentration states is created during a single cell generation than is created by cell division across a population. This persistence of molecular similarity down a lineage constitutes a form of heritable short-term memory. We identify the biochemical networks that produce, modify, and degrade molecules as an underexplored source of these additional molecular concentration states. Using experimentally informed simulations, we find that the strength and duration of molecular similarity down a lineage depend on tunable network properties, explaining why some cellular traits persist only briefly while others last generations. These contributions to molecular concentration variance from biochemical reaction networks act in concert with gene expression and other regulatory processes to shape the protein composition of cells. Our framework yields clear, testable predictions for determining how biochemical network architectures drive non-genetic cellular inheritance.
- Open Access
- 7 November 2025
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 122 (45)
Stability through plasticity: Finding robust memories through representational drift
Memories are believed to be stored in synapses and retrieved by reactivating neural ensembles. Learning alters synaptic weights, which can interfere with previously stored memories that share the same synapses, creating a trade-off between plasticity and stability. Interestingly, neural representations change even in stable environments, without apparent learning or forgetting—a phenomenon known as representational drift. Theoretical studies have suggested that multiple neural representations can correspond to a memory, with postlearning exploration of these representation solutions driving drift. However, it remains unclear whether representations explored through drift differ from those learned or offer unique advantages. Here, we show that representational drift uncovers noise-robust representations that are otherwise difficult to learn. We first define the nonlinear solution space manifold of synaptic weights for fixed input–output mappings, which allows us to disentangle drift from learning and forgetting and simulate drift as diffusion within this manifold. Solutions explored by drift have many inactive and saturated neurons, making them robust to weight perturbations due to noise or continual learning. Such solutions are prevalent and entropically favored by drift, but their lack of gradients makes them difficult to learn and nonconducive to future learning. To overcome this, we introduce an allocation procedure that selectively shifts representations for new stimuli into a learning-conducive regime. By combining allocation with drift, we resolve the trade-off between learnability and robustness.
- 7 November 2025
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 122 (45)
Open AccessStability through plasticity: Finding robust memories through representational drift
Memories are believed to be stored in synapses and retrieved by reactivating neural ensembles. Learning alters synaptic weights, which can interfere with previously stored memories that share the same synapses, creating a trade-off between plasticity and stability. Interestingly, neural representations change even in stable environments, without apparent learning or forgetting—a phenomenon known as representational drift. Theoretical studies have suggested that multiple neural representations can correspond to a memory, with postlearning exploration of these representation solutions driving drift. However, it remains unclear whether representations explored through drift differ from those learned or offer unique advantages. Here, we show that representational drift uncovers noise-robust representations that are otherwise difficult to learn. We first define the nonlinear solution space manifold of synaptic weights for fixed input–output mappings, which allows us to disentangle drift from learning and forgetting and simulate drift as diffusion within this manifold. Solutions explored by drift have many inactive and saturated neurons, making them robust to weight perturbations due to noise or continual learning. Such solutions are prevalent and entropically favored by drift, but their lack of gradients makes them difficult to learn and nonconducive to future learning. To overcome this, we introduce an allocation procedure that selectively shifts representations for new stimuli into a learning-conducive regime. By combining allocation with drift, we resolve the trade-off between learnability and robustness.
- Top Cited PapersOpen Access
- 12 February 2025
- Published by Springer Nature in Nature
- Vol. 640 (8057) , 165-175
Learning produces an orthogonalized state machine in the hippocampus
Cognitive maps confer animals with flexible intelligence by representing spatial, temporal and abstract relationships that can be used to shape thought, planning and behaviour. Cognitive maps have been observed in the hippocampus1, but their algorithmic form and learning mechanisms remain obscure. Here we used large-scale, longitudinal two-photon calcium imaging to record activity from thousands of neurons in the CA1 region of the hippocampus while mice learned to efficiently collect rewards from two subtly different linear tracks in virtual reality. Throughout learning, both animal behaviour and hippocampal neural activity progressed through multiple stages, gradually revealing improved task representation that mirrored improved behavioural efficiency. The learning process involved progressive decorrelations in initially similar hippocampal neural activity within and across tracks, ultimately resulting in orthogonalized representations resembling a state machine capturing the inherent structure of the task. This decorrelation process was driven by individual neurons acquiring task-state-specific responses (that is, ‘state cells’). Although various standard artificial neural networks did not naturally capture these dynamics, the clone-structured causal graph, a hidden Markov model variant, uniquely reproduced both the final orthogonalized states and the learning trajectory seen in animals. The observed cellular and population dynamics constrain the mechanisms underlying cognitive map formation in the hippocampus, pointing to hidden state inference as a fundamental computational principle, with implications for both biological and artificial intelligence.
- 12 February 2025
- Published by Springer Nature in Nature
- Vol. 640 (8057) , 165-175
Top Cited PapersOpen AccessLearning produces an orthogonalized state machine in the hippocampus
Cognitive maps confer animals with flexible intelligence by representing spatial, temporal and abstract relationships that can be used to shape thought, planning and behaviour. Cognitive maps have been observed in the hippocampus1, but their algorithmic form and learning mechanisms remain obscure. Here we used large-scale, longitudinal two-photon calcium imaging to record activity from thousands of neurons in the CA1 region of the hippocampus while mice learned to efficiently collect rewards from two subtly different linear tracks in virtual reality. Throughout learning, both animal behaviour and hippocampal neural activity progressed through multiple stages, gradually revealing improved task representation that mirrored improved behavioural efficiency. The learning process involved progressive decorrelations in initially similar hippocampal neural activity within and across tracks, ultimately resulting in orthogonalized representations resembling a state machine capturing the inherent structure of the task. This decorrelation process was driven by individual neurons acquiring task-state-specific responses (that is, ‘state cells’). Although various standard artificial neural networks did not naturally capture these dynamics, the clone-structured causal graph, a hidden Markov model variant, uniquely reproduced both the final orthogonalized states and the learning trajectory seen in animals. The observed cellular and population dynamics constrain the mechanisms underlying cognitive map formation in the hippocampus, pointing to hidden state inference as a fundamental computational principle, with implications for both biological and artificial intelligence.
- Open Access
Stability through plasticity: Finding robust memories through representational drift
Memories are believed to be stored in synapses and retrieved through the reactivation of neural ensembles. Learning alters synaptic weights, which can interfere with previously stored memories that share the same synapses, creating a tradeoff between plasticity and stability. Interestingly, neural representations exhibit significant dynamics, even in stable environments, without apparent learning or forgetting—a phenomenon known as representational drift. Theoretical studies have suggested that multiple neural representations can correspond to a memory, with post-learning exploration of these representation solutions driving drift. However, it remains unclear whether representations explored through drift differ from those learned or offer unique advantages. Here we show that representational drift uncovers noise-robust representations that are otherwise difficult to learn. We first define the non-linear solution space manifold of synaptic weights for a fixed input-output mapping, which allows us to disentangle drift from learning and forgetting and simulate representational drift as diffusion within this manifold. Solutions explored by drift have many inactive and saturated neurons, making them robust to weight perturbations due to noise or continual learning. Such solutions are prevalent and entropically favored by drift, but their lack of gradients makes them difficult to learn and non-conducive to further learning. To overcome this, we introduce an allocation procedure that selectively shifts representations for new information into a learning-conducive regime. By combining allocation with drift, we resolve the tradeoff between learnability and robustness.
Open AccessStability through plasticity: Finding robust memories through representational drift
Memories are believed to be stored in synapses and retrieved through the reactivation of neural ensembles. Learning alters synaptic weights, which can interfere with previously stored memories that share the same synapses, creating a tradeoff between plasticity and stability. Interestingly, neural representations exhibit significant dynamics, even in stable environments, without apparent learning or forgetting—a phenomenon known as representational drift. Theoretical studies have suggested that multiple neural representations can correspond to a memory, with post-learning exploration of these representation solutions driving drift. However, it remains unclear whether representations explored through drift differ from those learned or offer unique advantages. Here we show that representational drift uncovers noise-robust representations that are otherwise difficult to learn. We first define the non-linear solution space manifold of synaptic weights for a fixed input-output mapping, which allows us to disentangle drift from learning and forgetting and simulate representational drift as diffusion within this manifold. Solutions explored by drift have many inactive and saturated neurons, making them robust to weight perturbations due to noise or continual learning. Such solutions are prevalent and entropically favored by drift, but their lack of gradients makes them difficult to learn and non-conducive to further learning. To overcome this, we introduce an allocation procedure that selectively shifts representations for new information into a learning-conducive regime. By combining allocation with drift, we resolve the tradeoff between learnability and robustness.
- Open Access
From the fly connectome to exact ring attractor dynamics
A cognitive compass enabling spatial navigation requires the neural representation of head direction (HD), yet the neural circuit architecture enabling this representation remains unclear. While various network models have been proposed to explain HD systems, these models rely on simplified circuit architectures that may be irreconcilable with empirical observations from connectomes. Here we construct a neural network model for the fruit fly HD system that satisfies both connectome-derived architectural constraints and the functional requirement of continuous head-direction representation. To achieve this, we first characterized an ensemble of continuous attractor networks where compass neurons providing local mutual excitation are coupled to inhibitory neurons. Our multipopulation model allowed us to discover a new class of ring attractor network with weaker symmetry requirements than usually assumed. For each of the four available fly connectomes, our analyses uncovered three distinct realizations of these networks. Furthermore, we found that synaptic variations among and around the fly connectomes can be compensated by cell-type-specific rescaling of synaptic weights, which could be potentially achieved through neuromodulation. The fly connectomes are ideally configured to take advantage of this mechanism, suggesting a novel design principle linking synapse-resolution connectivity to network computation.
Open AccessFrom the fly connectome to exact ring attractor dynamics
A cognitive compass enabling spatial navigation requires the neural representation of head direction (HD), yet the neural circuit architecture enabling this representation remains unclear. While various network models have been proposed to explain HD systems, these models rely on simplified circuit architectures that may be irreconcilable with empirical observations from connectomes. Here we construct a neural network model for the fruit fly HD system that satisfies both connectome-derived architectural constraints and the functional requirement of continuous head-direction representation. To achieve this, we first characterized an ensemble of continuous attractor networks where compass neurons providing local mutual excitation are coupled to inhibitory neurons. Our multipopulation model allowed us to discover a new class of ring attractor network with weaker symmetry requirements than usually assumed. For each of the four available fly connectomes, our analyses uncovered three distinct realizations of these networks. Furthermore, we found that synaptic variations among and around the fly connectomes can be compensated by cell-type-specific rescaling of synaptic weights, which could be potentially achieved through neuromodulation. The fly connectomes are ideally configured to take advantage of this mechanism, suggesting a novel design principle linking synapse-resolution connectivity to network computation.
- Open Access
- 15 September 2024
- Published by Annual Reviews in Annual Review of Vision Science
- Vol. 10 (1) , 23-46
Optimization in Visual Motion Estimation
Sighted animals use visual signals to discern directional motion in their environment. Motion is not directly detected by visual neurons, and it must instead be computed from light signals that vary over space and time. This makes visual motion estimation a near universal neural computation, and decades of research have revealed much about the algorithms and mechanisms that generate directional signals. The idea that sensory systems are optimized for performance in natural environments has deeply impacted this research. In this article, we review the many ways that optimization has been used to quantitatively model visual motion estimation and reveal its underlying principles. We emphasize that no single optimization theory has dominated the literature. Instead, researchers have adeptly incorporated different computational demands and biological constraints that are pertinent to the specific brain system and animal model under study. The successes and failures of the resulting optimization models have thereby provided insights into how computational demands and biological constraints together shape neural computation.
- 15 September 2024
- Published by Annual Reviews in Annual Review of Vision Science
- Vol. 10 (1) , 23-46
Open AccessOptimization in Visual Motion Estimation
Sighted animals use visual signals to discern directional motion in their environment. Motion is not directly detected by visual neurons, and it must instead be computed from light signals that vary over space and time. This makes visual motion estimation a near universal neural computation, and decades of research have revealed much about the algorithms and mechanisms that generate directional signals. The idea that sensory systems are optimized for performance in natural environments has deeply impacted this research. In this article, we review the many ways that optimization has been used to quantitatively model visual motion estimation and reveal its underlying principles. We emphasize that no single optimization theory has dominated the literature. Instead, researchers have adeptly incorporated different computational demands and biological constraints that are pertinent to the specific brain system and animal model under study. The successes and failures of the resulting optimization models have thereby provided insights into how computational demands and biological constraints together shape neural computation.
- Open Access
Model-based inference of synaptic plasticity rules
Inferring the synaptic plasticity rules that govern learning in the brain is a key challenge in neuroscience. We present a novel computational method to infer these rules from experimental data, applicable to both neural and behavioral data. Our approach approximates plasticity rules using a parameterized function, employing either truncated Taylor series for theoretical interpretability or multilayer perceptrons. These plasticity parameters are optimized via gradient descent over entire trajectories to align closely with observed neural activity or behavioral learning dynamics. This method can uncover complex rules that induce long nonlinear time dependencies, particularly involving factors like postsynaptic activity and current synaptic weights. We validate our approach through simulations, successfully recovering established rules such as Oja’s, as well as more intricate plasticity rules with reward-modulated terms. We assess the robustness of our technique to noise and apply it to behavioral data from Drosophila in a probabilistic reward-learning experiment. Notably, our findings reveal an active forgetting component in reward learning in flies, improving predictive accuracy over previous models. This modeling framework offers a promising new avenue for elucidating the computational principles of synaptic plasticity and learning in the brain.
Open AccessModel-based inference of synaptic plasticity rules
Inferring the synaptic plasticity rules that govern learning in the brain is a key challenge in neuroscience. We present a novel computational method to infer these rules from experimental data, applicable to both neural and behavioral data. Our approach approximates plasticity rules using a parameterized function, employing either truncated Taylor series for theoretical interpretability or multilayer perceptrons. These plasticity parameters are optimized via gradient descent over entire trajectories to align closely with observed neural activity or behavioral learning dynamics. This method can uncover complex rules that induce long nonlinear time dependencies, particularly involving factors like postsynaptic activity and current synaptic weights. We validate our approach through simulations, successfully recovering established rules such as Oja’s, as well as more intricate plasticity rules with reward-modulated terms. We assess the robustness of our technique to noise and apply it to behavioral data from Drosophila in a probabilistic reward-learning experiment. Notably, our findings reveal an active forgetting component in reward learning in flies, improving predictive accuracy over previous models. This modeling framework offers a promising new avenue for elucidating the computational principles of synaptic plasticity and learning in the brain.